Investigating the spatial risk distribution of West Nile virus disease in birds and humans in southern Ontario from 2002 to 2005

BACKGROUND: The West Nile virus (WNv) became a veterinary public health concern in southern Ontario in 2001 and has continued to threaten public health. Wild bird mortality has been shown to be an indicator for tracking the geographic distribution of the WNv. The purpose of this study was to investigate the latent risk distribution of WNv disease among dead birds and humans in southern Ontario and to compare the spatial risk patterns for the period 2002–2005. The relationship between the mortality fraction in birds and incidence rate in humans was also investigated. METHODS: Choropleth maps were created to investigate the spatial variation in bird and human WNv risk for the public health units of southern Ontario. The data were smoothed by empirical Bayesian estimation before being mapped. Isopleth risk maps for both the bird and human data were created to identify high risk areas and to investigate the potential relationship between the WNv mortality fraction in birds and incidence rates in humans. This was carried out by the geostatistical prediction method of kriging. A Poisson regression analysis was used to model regional human WNv case counts as a function of the spatial coordinates in the east and north direction and the regional bird mortality fractions. The presence of disease clustering and the location of disease clusters were investigated by the spatial scan test. RESULTS: The isopleth risk maps exhibited high risk areas that were relatively constant from year to year. There was an overlap in the bird and human high risk areas, which occurred in the central-west and south-west areas of southern Ontario. The annual WNv cause-specific mortality fractions in birds for 2002 to 2005 were 31.9, 22.0, 19.2 and 25.2 positive birds per 100 birds tested, respectively. The annual human WNv incidence rates for 2002 to 2005 were 2.21, 0.76, 0.13 and 2.10 human cases per 100,000 population, respectively. The relative risk of human WNv disease was 0.72 times lower for a public health unit that was 100 km north of another public health unit. The relative risk of human WNv disease increased by the factor 1.44 with every 10 positive birds per 100 tested. The scan statistic detected disease cluster in the bird and human data. The human clusters were not significant, when the analysis was conditioned on the bird data. CONCLUSION: The study indicates a significant relationship between the spatial pattern of WNv risk in humans and birds.